Mufananidzo Ongororo: Chii Uye Mashandisiro Acho

by kim Updated: June 29, 2022

Ongororo yemifananidzo chishandiso chine simba chekuburitsa ruzivo kubva mumifananidzo.

Izvi zvinogona kushandiswa kune zvinangwa zvakasiyana-siyana, kubva pakutevera kuchinja kwezvakatipoteredza kusvika pakuonekwa kwechiso kusvika pakuongororwa kwemifananidzo yezvokurapa.

Muchikamu chino, tichave tichikupa sumo pfupi yekuongorora mifananidzo uye kuti ungaishandisa sei. Isu tinovhara tsananguro yekuongorora kwemifananidzo, maitiro akasiyana uye maturusi aripo, uye zvingangoita mashandisiro ekuongorora mifananidzo.

Tsanangudzo yekuongorora mufananidzo

Ongororo yemifananidzo chishandiso chinoshandiswa kuwana nzwisiso digitaalinen mifananidzo. Inowanzoshandiswa munzvimbo dzakasiyana siyana dzakadai semushonga, tsvakiridzo yesainzi, kuongorora, kugadzirisa mifananidzo, nezvimwe zvakawanda.

Chaizvoizvo, inzira yekuunganidza data kubva pamufananidzo uye kuongorora iyo data kuti uwane ruzivo nezve zviri pamufananidzo. Ongororo yemifananidzo inobatsira kupa ruzivo nezve mhando kana maficha echifananidzo senge saizi, chimiro, kusimba / kupenya, ruvara kuumbwa kana hurongwa hwayo hwepasi.

Kune akati wandei maitiro anoshandiswa mukuongorora mifananidzo senge spatial shanduko matekiniki senge sub-image kudhirowa uye block process; inoratidzira maitiro ekubvisa anowana mapoinzi kana matunhu anofarira uchishandisa edge detectors kana muganhu wekutsvaga algorithms; maitiro ekuronga anoisa zvinhu muzvikamu; segmentation maitiro anotsanangudza zvinhu kubva kumashure; uye maitiro ekuzivikanwa algorithms anoziva zvinhu uchishandisa nzira dzenhamba.

Nekushandisa maturusi ekuongorora mapikicha ezvishandiso zvakaita sekuona chinhu uye kuzivikanwa kwechiso, zvinogona kuvandudza ruzivo rwemushandisi kana uchishandisa dhijitari media. Zvinogoneka zvakare kushandisa ongororo yetsvagurudzo yesainzi kuti uongorore kuwirirana pakati pemiganhu yakatsanangurwa mukati meimwe kana boka remifananidzo. Chekupedzisira, tekinoroji iyi inoshandiswawo zvakanyanya nenyanzvi dzekurapa pahunyanzvi hunosanganisira radiology uye pathology kuongorora mamiriro akasiyana zvichienderana nemifananidzo yakatorwa kubva kuvarwere.

Mhando dzekuongorora mifananidzo

Ongororo yemifananidzo ndiko kutorwa kweruzivo rwakakosha kubva kumifananidzo yedhijitari. Iri izwi rakatambanuka kwazvo rinobata akawanda akasiyana marudzi emabasa ane chekuita nemufananidzo, anosanganisira kuzivikanwa kwechinhu uye chiziviso, kuyerwa kwechimiro, kucherechedzwa kwepateni, kuongororwa kwenzvimbo uye kuronga.

Kune marudzi akawanda emifananidzo yekuongorora maitiro; zvimwe zvezvinonyanya kuzivikanwa zvinosanganisira:
-Pixel-based/Spatial Analysis - Iyi mhando yekuongorora inosanganisira kuverenga mapikisi ega ega kana masumbu emapikseli kuyera nzvimbo (saizi, chimiro) uye densities (kugovera).
-Statistical Analysis - Iyi nzira inoshandisa nzira dzenhamba yekuongorora maitiro ane hukama nematunhu mukati memufananidzo akadai sekusiyanisa kana kupenya.
-Machine Kudzidza / Artificial Intelligence - Artificial intelligence (AI) algorithms inogonesa ane simba otomatiki ekugadzirisa mifananidzo mabasa akadai sekuona chinhu kana semantic segmentation uchishandisa convolutional neural network (CNNs).
-Texture Analysis - Iyi nzira inoyera kuti zvinhu zviri mukati memufananidzo zvinogoverwa sei kune imwe neimwe maererano nekugadzirwa-kusiyana kwegrey mazinga kana mavara pamusoro penzvimbo yakakura.
-Histogram Analysis - Kuongorora kweHistogram kunowanzobatanidza kuyera kugoverwa kwenhamba mukati memufananidzo wakadai sekusiyanisa, kupenya, hue uye saturation.
-Kusefa Techniques - Masefa maitiro anosanganisira mashandiro emasvomhu pane yese yakagadziridzwa intensity array iyo inogadzirisa zvimwe zvikamu zveiyo array ichichengetedza zvimwe. Mafirita anowanzo shandiswa kuwedzera mipendero kana kubvisa ruzha kubva pamufananidzo nekutsvedzerera panzvimbo idzi dzisingadiwe.
-Matekinoroji mana ekushandura - Iyi nzira inoshandisa shanduko yeFourier iyo inoodza mifananidzo kuita zvikamu zvakati wandei zvichibva pane frequency renji uye/kana matarisiro aripo mune yekuisa furemu. Izvi zvinobvumira kupatsanura zvikamu uye zvinhu zvinogona kunge zvine humwe hunhu hunoenderana nezvimwe zvikamu zvakazara kutaridzika mufuremu - senge magadzirirwo kana misiyano yechimiro - zvichiita kuti zvive nyore kuona chinhu panguva yezvikamu.

Zvakanakira Kuongororwa Kwemifananidzo

Ongororo yemifananidzo chishandiso chakakosha chekuwana ruzivo kubva kumifananidzo yedhijitari. Inoshandiswa kuziva mapatani, kuona zvinhu, uye kubvisa ruzivo rwakakosha kubva munhau. Kubva pakutsvakurudza kwezvokurapa kusvika pakuonekwa kwechiso, kuongororwa kwemifananidzo kunogona kubatsira maindasitiri akawanda. Pano tichaongorora zvakanakira ongororo yemifananidzo uye kuti ingashandiswa sei kusimudzira akasiyana siyana.

Yakagadziridzwa ruzivo rwevatengi

Ongororo yemifananidzo inogona kubatsira kuvandudza zviitiko zvevatengi, zvichibvumira makambani kuti anzwisise zvinodiwa nevatengi nekukurumidza uye nenzira kwayo. Nekushandisa nzira dzekuongorora mifananidzo, mabhizinesi anogona kushandisa data kubva kumifananidzo yakadai sehuwandu hwevanhu, nzvimbo, uye mhinduro dzemanzwiro kuti vatarise kushambadzira kwavo kana kugadzirisa zvemukati kune vatengi. Kushandisa iyi data kunogona kutungamira mukudyidzana nevatengi pamwe nekuwedzera kutengesa kwekambani.

Ongororo yemifananidzo inobatsirawo mabhizinesi kunzwisisa zviri nani mamiriro emifananidzo yavo. Semuenzaniso, tekinoroji yekuziva mavara inobvumira makambani kuona mavara mumufananidzo uye kuona kuti anodyidzana sei nezvigadzirwa kana masevhisi. Ruzivo urwu runogona kubatsira mabhizinesi kuwana kunzwisisa mumaitiro evatengi maererano nesevhisi kana chigadzirwa. Pamusoro pezvo, mabhizinesi anogona kushandisa tekinoroji yekuziva zvinhu kuona zvinhu zviri mumufananidzo uye kunzwisisa zviri nani zvinodiwa nevatengi zvichibva pazvinhu zvavanoda.

Chekupedzisira, ongororo yemifananidzo inoita kuti mabhizinesi atore data nekukurumidza achiri kuchengetedza huchokwadi mukuita. Makambani haachadi kuongorora nemaoko mazana emifananidzo paunenge uchiedza kuona zvinodiwa nevatengi; asi ivo vanokwanisa kushandura maitiro aya vachishandisa nzira dzekuongorora mifananidzo iyo inovandudza kushanda kwayo zvakanyanya vachiri kuwana mhedzisiro kubva pakuongorora kwayo kwemifananidzo. Izvi zvinoreva kuti vatengi vanowana mukana wekushambadza kwemunhu nekukurumidza kupfuura nakare kose, zvichiita kuti makambani ape zvirinani mushandisi ruzivo kune vatengi vavo.

Kuwedzera kururama uye kushanda zvakanaka

Ongororo yemifananidzo inopa akawanda mabhenefiti kumabhizinesi, kusanganisira kuwedzera kurongeka uye kushanda nesimba. Nekuongorora mifananidzo pachinzvimbo chekuvimba nemanyorero kana zvinyorwa-zvakavakirwa nzira, kuongororwa kwemifananidzo kunogona kuderedza zvakanyanya kukanganisa kwekupinda data uye kukurumidza kugadzirisa nguva yebasa rakapihwa.

Ongororo yemifananidzo inobvisawo kukosha kwekusefa mukati mehuwandu hwemashoko nemaoko. Pane kudaro, algorithm inogona kukurumidza kuwana data rakakodzera uye kuona mapatani kuti adudzirwe zviri nyore nevanhu. Izvi zvinomhanyisa maitiro ekuita sarudzo uye zvinoderedza zvingangodhura kukanganisa nekuda kwekukanganisa kwevanhu kana kutarisa.

Pamusoro pezvo, ongororo yemifananidzo inoita kuti zvive nyore kuziva mapatani nekukurumidza mumaseti makuru akakura izvo zvingatora zviuru zvemaawa evarume kupepeta nemaoko. Izvi zvinoderedza zvakanyanya mutengo wevashandi uku uchiwedzera huchokwadi uye kushanda zvakanaka panguva imwe chete. Ongororo yemifananidzo inobvumirawo masangano kuti averenge maitiro aangave akapotsa nenzira dzechinyakare, zvichivabatsira kuita sarudzo dziri nani nekukurumidza kupfuura nakare kose.
Nekushandisa tekinoroji yekuongorora mifananidzo, mabhizinesi anogona kuderedza zvikanganiso zvekupinda kwedata uye kuderedza mutengo wevashandi uku achiwedzera huchokwadi uye kugona mukushanda kwavo. Izvi zvinotungamira mukuvandudzika kwekuita sarudzo uye mhedzisiro dzakanyatsojeka dzinotorwa kubva kune yakakura kwazvo dataset pane izvo zvinogona kuwanikwa uchishandisa nzira dzechinyakare chete.

Kuvandudzwa kwekuita zvisarudzo

Kushandiswa kwekuongorora kwemifananidzo mumabhizinesi kubatsira nekuita sarudzo kunogona kutungamira mukuvandudza kushanda uye kubatsira masangano kutonga zvirinani zviwanikwa zvavo. Kuburikidza nemaitiro akasiyana uye matekinoroji anosanganisirwa, sangano rinokwanisa kukurumidza kugadzirisa uye kuongorora mifananidzo kubva kwakasiyana zvinyorwa. Izvi zvinovabatsira kuita zvisarudzo zvine musoro ipapo ipapo.

Ongororo yemifananidzo inogona kubatsira masangano kuronga, kuona, uye kuongorora data kuitira kukoshesa mabasa kana zvinangwa zvine hungwaru. Inobvumirawo kukurumidza kuongororwa kwemaitiro akaoma mukati memifananidzo, iyo inochengetedza nguva yakakosha pamabasa epamusoro-soro ekuita sarudzo. Pamusoro pezvo, kuongororwa kwemifananidzo kunogona kubatsira makambani kuwana mafambiro kana nyaya dzinodzokororwa dzinogona kunge dzisina kucherechedzwa.

Masangano ari kuwedzera kuwedzera simba rekuongorora mifananidzo senge artificial intelligence (AI) uye muchina kudzidza (ML) maapplication kugadzira mamodhi ekufungidzira anovapa muono wezvichaitika maererano nemaitiro evatengi uye mamiriro emusika mune ramangwana. Panguva ino, zvikwata zvekuongorora zvemusika zvinogara zvichitarisa kuburikidza nemifananidzo yavakaunganidza kubva kuzvinyorwa zvakaita seanokwikwidza mawebhusaiti uye nzira dzedhijitari. Nematurusi akakodzera, zvikwata izvi zvinogona kukurumidza kuona kuti vatengi vanodyidzana sei nezvigadzirwa kana masevhisi esangano kuitira kuvaka zvinobatika zvinopa mukana wemakwikwi pamusoro pevamwe muindasitiri.

Ongororo yemifananidzo inoita basa rakakosha mukushanda kwebhizinesi nhasi nekupa masangano mukana wekuwana ruzivo rwakakosha runofambiswa nedata rwunokurudzira kukura nekufambira mberi nekukurumidza kupfuura nakare kose.

Mashandisiro Akaitwa Kuongorora Mifananidzo

Ongororo yemifananidzo inzira inoshanda yekushandisa yakawanda data data uye kubvisa ruzivo rwakakosha kubva mairi. Ongororo yemifananidzo inosanganisira kushandiswa kwemacomputer algorithms kuona zvinorehwa nemufananidzo. Inogona kushandiswa kune akasiyana mabasa akadai sekuziva chinhu, kuzivikanwa kwechiso, uye kupatsanurwa kwemifananidzo. Muchikamu chino, tichakurukura kuti ongororo yemifananidzo inogona kushandiswa sei uye mamwe mashandisirwo ayo.

Image segmentation

Chikamu chemifananidzo ibazi rekugadzirisa mufananidzo umo mufananidzo wakakamurwa kuita zvikamu zvakawanda, kana matunhu, ayo pamwechete anoumba mufananidzo wacho wese. Chikamu chega chega chinomiririra nzvimbo yekufarira mukati memufananidzo, uye inogona kufungidzirwa sechinhu chakasarudzika pachayo. Chinangwa chekupatsanura ndechekurerutsa uye/kana kushandura kumiririrwa kwechifananidzo kuita chimwe chinhu chine musoro uye chiri nyore kuongorora. Segmentation matekiniki anoshandiswa mukufungidzira kwekurapa kweakasiyana mabasa, kubva pakuziva anatomical zvimiro kusvika pakutarisa kufambira mberi kwechirwere. Uye zvakare, segmentation algorithms zvinhu zvakakosha mune otomatiki yekutarisa masisitimu uye inogona kushandiswa kune marobhoti navigation mabasa uye kuona chinhu.

Segmentation algorithms kazhinji inotsamira pane chero pixel-yakavakirwa nzira kana kupatsanurwa-kwakavakirwa nzira. Pixel-based nzira dzinoshandisa zviratidzo zvakaita seruvara kana magadzirirwo kupatsanura mufananidzo kuita zvikamu zvakasiyana. Nerumwewo, kupatsanurwa-kwakavakirwa nzira dzinobatanidza zvikamu zvinoenderana nezvinotsanangudza zvakaita sechimiro kana magadzirirwo kuitira kuti otomatiki mapoka mapixels pamwe chete muzvikamu/zvinhu zvakakodzera.

Pixel-based segmentation inowanzovavarira kupatsanura zvinhu nehukuru hwazvo kana mutsauko nekuverenga nhamba dzepasirese (semuenzaniso, kukosha) munzvimbo yese. Mutsara-wakavakirwa nzira dzinowanzoda manyorerwo ekuisa uye kunyora mazita algorithm isati yatanga kushanda; zvisinei nzira idzi dzine simba zvikuru pakubata nedatasets hombe kana marudzi asingazivikanwe data ayo asingakwanisi kupatsanurwa zvinobudirira uchishandisa chete pixel values ​​chete.

Kuronga kwemifananidzo

Kurongeka kwemapikicha maitiro ekushandisa komputa algorithms kufanotaura uye kuisa makirasi mumufananidzo wakapihwa. Mukuita uku, zvakajairika kuve ne "muunganidzwa" wemifananidzo yekudzidziswa yakanyorwa iyo inoshanda seyekupinza mune yekudzidza algorithm. Zvichienderana nemifananidzo iyi yekudzidzisa, maficha akadai saizi, chimiro uye ruvara anotorwa kubva pamufananidzo uye anoshandiswa kurongedza mufananidzo. Iyi nzira inobvumira mibairo yechokwadi kana ichienzaniswa nekunyora kwemanyorero, uko vanhu vanogona kukanganisa nekuda kwekukanganisa kana kukanganisa kwemazita.

Tichipihwa mufananidzo usingazivikanwe, munhu anogona kushandisa iyo iripo classifier iyo inozopa iyi chaiyo yekuisa chikamu chekubuda zvinoenderana nenhamba yakafanotaurwa yemakirasi akasiyana. Mimwe mienzaniso yekushandiswa kweiyi inotangira kubva kune yakazvimiririra yekutyaira mota ichipa kugona kwekuziva chinhu kusvika pakuonekwa kwechiso mumashandisirwo ebiometrics. Pamusoro pezvo, kuve nedata rakanyorwa rakanyorwa kuburikidza nekuiswa kwemifananidzo kunogona kupa ruzivo rwakanyanya munharaunda yedu nekushandisa ma dataseti makuru ehurumende kuti awedzere kuongororwa - kunyanya muhukama pakati pemhando dzezvinhu uye mashandisirwo ebhizinesi akadai sekuongorora kwekutengesa kana ongororo yevaraidzo.

Kubva pakushandisa neural network neGPU optimization matekiniki kuitira kuti uwane parallel process kusvika kune dzakasiyana siyana dzekudzidza muchina - kune nzira dzisingaverengeke munhu anogona kudzidzisa maAI mamodheru kana kushandisa akadzama ekudzidza mumunda weImage Classification; kunyangwe kudzidza kusingatarisirwe kuchine zvimwe zvipingamupinyi zvinoita kuti vaongorori vasazvishandise zvizere pakugona kukuru, nzira idzi dzichiri kunyatsoongororwa nekuferefetwa. Saka kunzwisisa mashandisirwo aya kunoda kubata kwakadzama pamacomputer echiono algorithms pamwe chete neruzivo rwakasimba rwehunyanzvi runosanganisira maitiro einjiniya ekumashure senge mitauro yekunyora sePython yezvinangwa zvekushandisa pamwe nekuva neruzivo rwepedyo nezve tekinoroji iri kusimukira senge blockchain kana server-shoma zvivakwa. kushandisa

Kuonekwa kwechinhu

Kuonekwa kwechinhu tekinoroji yakavakirwa pakombuta uye hunyanzvi hwekuona huchokwadi mukuongororwa kwedataset yemufananidzo. Iyi tekinoroji inoshandisa akasiyana sophisticated algorithms kuona zvinhu uye hunhu hwazvo mumifananidzo inoonekwa zviri nyore. Inobata basa rakakura munzvimbo dzakawanda, senge forensics, sainzi yekurapa, maindasitiri otomatiki, kuzivikanwa kwechiso uye otomatiki ekuongorora masisitimu.

Kuonekwa kwechinhu kunosanganisira kuongorora kana kuunganidza dhata kuti uone ukuru, chimiro uye zvimwe zvinhu zvechimwe chinhu chiri mumufananidzo. Iyi mhando yekuongorora mifananidzo inogona kusanganisira kuziva zvinhu zvichienderana neruvara rwazvo, magadzirirwo kana kunyange maumbirwo azvinosika kana akaiswa pamwechete. Kushandisa masvomhu uye nzira dzechiverengero kunoita kuti basa rive nyore kumakombuta mazuvano.

Chinangwa chikuru apa ndechekuona chinhu chega chega chinoonekwa mumufananidzo wedataset nekushandisa iwo akapihwa hunhu hwemavara. Iyo algorithm inoda kudzidziswa kare nedata rakawanda, kusanganisira 'yakanaka' mienzaniso yekufungidzira kwakaringana nezvechinhu chimwe nechimwe chinonangwa kudzidziswa kusati kwatanga. Mushure mekunge kudzidziswa kwapera uye kufembera kwabudirira, kunoteverwa nezvikamu senge inference apo mhedzisiro inotarisirwa kubva kune yakapihwa inoshanda nemhedzisiro yakawanikwa kubva kuzvidzidzo zvekare.

Tekinoroji senge kudzidza kwakadzama (DL) maalgorithms anobatsira nekubvumidza akawanda akaturikidzana ayo anoshanda pamwe chete kuti awane mhedzisiro chaiyo mukati menguva pfupi yenguva kupfuura nzira dzechinyakare dzaigona kupa isati yave kuwanikwa mukati merudzi urwu rwekugadzirisa Hardware setup inoshandiswa nhasi. Pakupedzisira izvi zvinoita kuti nzira dzeAI dzinoyamurwa seKudzidza Kwakadzika kuwedzere kutaridzika kupfuura nzira dzakajairwa sezvo ivo vachideredza zvakanyanya zvikanganiso vachipa huchokwadi hwekuona kwechinhu metrics inogona kuburitsa zviri nyore kuburitsa mune chaiyo nguva mamiriro zvakare.

Zvishandiso zveKuongorora Mifananidzo

Ongororo yemifananidzo inzira inosanganisira kutora ruzivo rwakakosha kubva pamifananidzo nemaitiro ayo ekombuta. Iyi nzira inoshandiswa munzvimbo dzakasiyana-siyana, dzakadai sekufungidzira kwechiremba, njere dzekugadzira, uye kure kure. Nenzira yakanaka, kune zvakasiyana-siyana zvezvishandiso zviripo zvekuongorora mufananidzo, zvichiita kuti hurongwa huve nyore uye hunobudirira. Ngatitarisei zvimwe zvezvishandiso zviripo pakuongorora mifananidzo.

OpenCV

OpenCV iraibhurari yemabasa ekuronga anonyanya kunangwa pane chaiyo-nguva komputa kuona. Inotsigira akasiyana mahwindo, Linux, chikuva uye mamwe mapuratifomu esoftware uye saka inowanzoshandiswa pakuongorora mifananidzo uye mabasa ekugadzirisa mifananidzo. Nzvimbo dzekushandisa dzeOpenCV dzinosanganisira: kuona komputa, kuzivikanwa kwechiso, kuzivikanwa kwechinhu, kutevedza mafambiro, kupatsanura uye kuzivikanwa.

OpenCV inopa huwandu hwakawanda hwezvishandiso zvine simba zvinogona kushandiswa kuita mabasa akaomarara sekuona chinhu, kuzivikanwa kwechiso, otomatiki ficha kudhirowa nezvimwe. Iine nharaunda inoshanda inopa mushandisi-ane hushamwari mhinduro kumatambudziko akajairika. Raibhurari iyi yakagadzirirwa kubatanidza nemitauro yakawanda yekuronga seC ++, Java kana Python. Raibhurari inogonesa vanogadzira kuti vashande nemifananidzo mumitauro yavo yavanoda nokudaro kuderedza nguva yebudiriro zvakanyanya. OpenCV inotsigira chero mhando yedata (senge mifananidzo iri mu2D uye 3D kuumbwa) iyo inoita kuti ive chishandiso chinonyanya kushandiswa pakati pemaraibhurari ese ekugadzirisa mifananidzo aripo nhasi.

Zvishandiso zvinosanganisirwa muOpenCV ndezvi:
-Image kusefa
-Object segmentation
-Features kutorwa
-Kuenzanirana kwemaitiro uye kupatsanura
-Object tracking uye kufambisa kuongorora
-3D kuvakwazve kwezvinhu zviri kudzidza

Zvishandiso zvaro zvinoshandiswa zvakanyanya kuita otomatiki yekuona yekuongorora mumaindasitiri nekuda kweayo anobatsira maficha senge yakaderera ndangariro kushandiswa, kukurumidza kuuraya pamapuratifomu mazhinji (kunyanya akaiswa mukati), kutakurika pakati pemasisitimu anoshandisa akajairwa maraibhurari seOpenMPI/MPI4Py. Rezinesi rayo rakavhurika sosi rinoita kuti iwedzere kutaridzika kune vashandisi vanoda kuishandisa pasina mubhadharo wehumambo kana marezinesi kubva kumakambani kana masangano akaita semasvomhu nezvimwe…

TensorFlow

TensorFlow iraibhurari yakavhurwa sosi software inoshandiswa kugadzira nekudzidzisa mamodheru ekudzidza akadzama. Masisitimu ekudzidza akadzama, akadai semamodhi ekuongorora mifananidzo, anovakwa neakawanda akaturikidzana eeuroni anogadzirisa data rekuisa, semufananidzo. TensorFlow inogonesa vanogadzira kuti vagadzirise zviri nyore uye kuendesa mamiriro-e-iyo-yakadzika maitiro ekudzidza anogona kushandiswa muchiratidzo, kugadzirwa kwemutauro wechisikigo nedzimwe nzvimbo.

Kubatsira kukuru kwekushandisa TensorFlow ndeyekuti inopa nzira iri nyore yekugadzira uye kuendesa ane simba emuchina ekudzidza modhi nekukurumidza uye nemazvo. TensorFlow inotsigirawo akati wandei maapplication mukati meMacOS, Windows, Linux uye nharembozha. Uye nekuti yakavhurika sosi uye ine hupfumi hwerutsigiro rwenharaunda, unogona kushandisa maturusi aripo mukuita kwako kwekusimudzira pasina kudzoreredza mavhiri kana kushandisa imwe nguva pakugadzirisa kodhi kubva kutanga.

Pamusoro pekudzidzisa mamodheru matsva kubva kutanga, TensorFlow zvakare inobvumidza iwe kuti ushandise zvekare-yakadzidziswa mamodheru ekufambisa kudzidza kana kugadzirisa zvakanaka pane mamwe matambudziko seti. Izvi zvinoita kuti zvikwanise kukurumidza kuvaka zvigadziriso zvakagadzirirwa kune mamwe matambudziko uchitora mukana weruzivo rwuripo rwakavakwa mumatiweki aripo. Nekuchinjika uku kuri muruoko, vagadziri vanogona kutanga kuyedza zvivakwa zvakasiyana nekukurumidza kupfuura nakare kose - kuasimudza uye kumhanya nekurongeka kwepamusoro kunoguma pasina nguva zvachose.

Google Cloud Vision

Google Cloud Vision chishandiso chine simba chekugadzirisa mifananidzo chakagadzirirwa kubatsira mabhizinesi kuongorora mifananidzo uye kuwana ruzivo rwakakosha. Iyo yakavakirwa pane imwecheteyo tekinoroji inoshandiswa neGoogle Photos, ichiita kuti ive yakavimbika uye nyore kushandisa. Ichi chishandiso-chakavakirwa gore chinobvumira mabhizinesi kugadzirisa yakawanda yemifananidzo nekukurumidza, kuona zviso, kuona zvinhu, uye kubvisa zvinyorwa zvese uku uchinzvenga basa remaoko.

Chimwe chinhu chikuru cheGoogle Cloud Vision kugona kwayo kuona zvinhu kubva mumhando dzakasiyana dzemhando semhuka kana zvigadzirwa. Iine "logo yekuona" ficha, vashandisi vanogona kubatana nemhando seFacebook kana Instagram kune yakadzama yekutevera uye yekuongorora ruzivo. Iyo "nzvimbo yekuona nzvimbo" inobatsira vashandisi kuona nzvimbo dzakavakomberedza uye kunzwisisa tsika dzekwavo zvirinani nekukurumidza.

Kunze kwekugona kucherechedzwa kwechinhu, Google Cloud Vision inopawo vashandisi kuongororwa kwemanzwiro pamwe neautomated captioning chishandiso chemifananidzo - yakanakira kuongororwa ipapo! Pamusoro pezvo, chishandiso ichi chinopa kugona kucherechedzwa kwechiso izvo zvakanakira kushambadzira sezvo ivo vachiona hunhu hwakadai sezera uye murume kana mukadzi nekurongeka kwepamusoro. Chekupedzisira, iwo akakosha ekubvisa masimba anogonesa vashandisi kukurumidza kuongorora kuburikidza nekuisa mifananidzo vachiwana akakosha mazwi anogona kushandiswa mune gare gare maitiro kana maapplication.

Mukupedzisa, Google Cloud Vision isarudzo yakanaka kana uchitsvaga chishandiso chekugadzirisa mifananidzo kubatsira mabhizinesi kuongorora mifananidzo yavo zvakanyanya kupfuura nakare kose!

mhedziso

Ongororo yemifananidzo chishandiso chine simba chekuunganidza zviono kubva mumifananidzo yedhijitari. Inogona kushandiswa munzvimbo dzakawanda dzakasiyana semushonga, mota, uye kutora mifananidzo. Mabhenefiti ekushandisa ongororo yemifananidzo akawanda uye anogona kuchengetedza nguva, mari, uye kushanda nesimba. Muchikamu chino, takatarisa nheyo dzekuongorora kwemifananidzo, kuti chii, uye kuti ingakubatsira sei. Takakurukurawo marudzi akasiyana-siyana ekuongorora uye nzira dzakanakisisa dzekubudirira kwekuongorora mufananidzo. Mukupedzisa, zviri pachena kuti ongororo yemifananidzo chishandiso chine simba chekuunganidza ruzivo uye kuita sarudzo dzinotungamirwa nedata.

pfupiso

Sarudzo yekugocha yekofi yako inoenderana nekunaka uye simba raunosarudza pamwe nezvaunofarira zvedunhu. Vanogocha vakasiyana vanopa huwandu hwakasiyana hwemazita nemhando dzemavara, asi kazhinji, zvinyoro zvinyoro zvichava nekunakirwa kwakapfava uye zvishoma zvakakwirira zvine caffeine yemukati nepo rima rakagochwa richivava nekushomeka kushoma. Mhando dzakakurumbira dzeimwe neimwe dzinosanganisira Chiedza Guta, Hafu Guta, Cinnamon, Guta, American, Chisvusvuro, Guta Rakazara, High Continental New Orleans European Espresso Viennese Italian French. Kunyangwe iwe uri fan yechiedza kana rima bhinzi yekofi-kana kumwe pakati-ita shuwa kubvunza usati watenga kuti uwane chaizvo zvauri kutsvaga!

Kuverenga Zvakare

Kushandiswa kwekuongorora kwemifananidzo kuti uwane nzwisiso uye kutora mhedziso kubva pamifananidzo iri kuramba ichizivikanwa mumaindasitiri akasiyana-siyana, kubva kuhutano kusvika kune kuchengeteka. Kana iwe uchida kudzidza zvakawanda nezve tekinoroji iyi, funga zvinotevera zviwanikwa:

Zvinyorwa & Tutorials
- "Mufananidzo Wekuongorora: Chii uye Mashandisiro Acho" naOracle
- "Nheyo yeKuongorora Kwemifananidzo Kune Vanhu Vasingazive Chinhu Nezve Kuongororwa Kwemifananidzo" naKent Woodard
-"Matanho masere eKunzwisisa Ongororo Yemifananidzo uye Kugadziridza muKudzidza kweMichina" naVictor Charpenay
Mabhuku & Mabhuku
-Image Processing: Misimboti uye Zvishandiso naMilan Sonka et al.
-Kombuta uye muchina Vision - Theory, algorithms, Practicalities (4th Ed.) naER Davies.
-Digital Image Kugadziriswa neOpenCV - Bibliography (Ver 4.1) Yakagadzirwa naDavid Dardas Webinars & Podcasts
-Computer Vision Kuburikidza nemuchina Kudzidza Webinar inotungamirwa naTreehouse AI
-Machine Learning Fundamentals Podcast hosted by Google Cloud Platform Kunewo mamwe mabhuku akawanda, zvinyorwa, webinars, podcasts, workshops nezvimwe zviwanikwa zviripo zvinogona kukubatsira kuti unzwisise zviri nani nzira yekuongorora mifananidzo pamwe nemashandisirwo ayo kune chaiwo maapplication.

Mhoroi, ndini Kim, amai uye munhu anofarira zvekumira-famba ane ruzivo mukugadzira midhiya nekuvandudzwa kwewebhu. Ndine chido chikuru chekudhirowa uye mifananidzo, uye ikozvino ndiri kunyura nemusoro munyika yekumira-kufamba. Neblog rangu, ndiri kugovera zvandinodzidza nemi varume.